AI News

Turn One Photo Into a ChatGPT Sticker Pack

A practical ChatGPT Images workflow for turning one approved portrait or original mascot into a coherent eight-reaction sticker set people can recognize and use.

Grok Imagine’s Quality API model retires Nov. 2.

xAI will redirect the old image-model slug to Image 2.0 at low quality. Existing requests keep running, but output, model reporting, and price can change underneath the same prompt.

Give Grok a Copy Map First

SpaceXAI presents Grok Imagine Image 2.0 as a model for dense layouts and sharp small text. Give it a literal copy map before asking for style.

Ground Gemini Images Before You Style Them

Nano Banana 2 can use Google Search grounding with Web and Image Search. Separate the traits that must be true from the mood the final frame should carry.

Midjourney takes edit prompts.

The first V8.2 edit model opened to every Midjourney account on August 27. Instruction edits, four reference images, inpainting and outpaint in one place. What changes is which prompt you write first.

A new tool turns rough video into 4K.

Black Forest Labs released a video upscaler on 20 August. You can now make many low cost takes and pay for quality once, at the end. One setting is on by default, and it quietly changes faces.

The prompts are worth more than the verdict.

Recraft put seven briefs through its own V4.1 and through Ideogram V4, then told us which one won. A company marking its own exam is not evidence. The seven briefs, printed in full, are a different matter entirely.

Sora has a shutdown date.

The app and the website went dark on the twenty-sixth of April. Everything that is left switches off on the twenty-fourth of September. On the table where OpenAI lists what to move to, the row for Sora holds a dash.

The prompt stopped being text.

Google switched off three Imagen 4 endpoints today. What replaces them has no separate image call at all: pictures now come out of the same request as words, and that request will accept fourteen reference images, a video, or a link to one.

The new rate has a date on it.

Google shipped a workhorse model on the thirteenth of August and put an introductory number on it: three quarters of a dollar per million tokens in. That number holds until the thirty-first of December. On the first of January it doubles.

The download became the product.

Suno shipped a studio, signed with BMG, and put a meter on downloads, all in one week. The meter starts September 3.

The licence that finally says yes.

Meta published a 30 billion parameter multimodal model today, small enough for one consumer card. The size will get the headlines. The three words that matter for paid work are Apache two point zero.

The unlimited plan has a date on it.

Runway's Unlimited plan runs until August 31 and becomes Max on September 1. If you pay annually, your decision is due this month, and what goes away is the part that let you look before you knew.

The model moved into 2.5 gigabytes.

Liquid AI put a 2.6 billion parameter model on Hugging Face on August 4 that runs in under 2.5 GB of memory. It makes no pictures at all, which is exactly why it is worth your attention this week.

The first courtroom answered.

Shipped, or merely announced.

Four releases in seven days. The question worth asking is not which is most impressive. It is which one you can put to work this afternoon, and the answers are further apart than the headlines suggest.